Violence within the American family is a significant and pervasive problem. With rare exceptions, however, the empirical literature on partner abuse is distinct and separate from research on child abuse. Clinical services for these two forms of family violence are similarly non-overlapping. Even so, parent and partner aggression co-occur more often than predicted by base rates, and many of the risk factors identified in one dyadic domain have documented parallels in the other. The parallels suggest the possibility of common predictors for the two types of family violence, predictors which, because of their generality, could be potentially important targets for intervention. The overarching purpose of the current project is to bring to bear both the spouse and child abuse literatures in exploring family violence at the level of the family and in evaluation predictors of the full range of partner and parent aggression as exhibited by both men and women in the same families. We have identified variables likely to function as common risk factors for family aggression, in that they may simultaneously predict both partner and parent aggression. Further, we hypothesize that some risk factors are role-specific and predict only one form of aggression (e.g., jealously should predict partner, but not parent, aggression). Using established measures where possible and conducting parallel analyses for husbands/fathers and wives/mothers throughout, our goals are to (1) establish the predictive utility of a multivariate model for partner aggression; (2) establish the predictive utility of a multivariate model of parent aggression; (3) determine whether common variables simultaneously predict both partner and parent aggression; (4) determine whether role-specific variables predict within, but not across, domains of aggression; (5) assess the generalizability of the separate and common models to aggressive samples and to the prediction of injury; (6) determine the occurrence and co-occurrence of the four forms of dyadic aggression and the associations between different constellations of family violence and overall family adjustment; (7) explore the relation between the extent of aggression in one dyad and the extent of aggression in the others; and (8) explore the aggressive participants' views of the relations between episodes of partner and parent aggression. The study uses random digit dialing to solicit a representative sample of 400 couples who parent a 3-7 year-old child. Additional couples who meet the same criteria and in which one partner reports some level of mild partner and parent aggression will be solicited with the same procedure and will be combined with dually aggressive individuals from the representative sample to permit validation of the models within aggressive samples.